AI16 October 202610 min
The AEO Checklist for Technical Content
A 5-step AEO checklist for technical content. Structure tutorials and docs so ChatGPT, Perplexity, and Google AI Overviews cite them.
Theodore Kelechukwu OnyejiakuTechnical content marketing
Technical content developers read and AI cites.
We help devtool companies publish content built for search, answer engines, and AI models — SEO, AEO, and GEO. Tutorials, documentation, comparison and integration pages, and video shorts.
Our team has created content for
Strapi
The open-source headless CMS. A Node.js and TypeScript content API that teams either self-host or run on Strapi Cloud.
strapi.io
OpenReplay
Open-source session replay and product analytics, built to be self-hosted so the recordings never leave your own infrastructure.
openreplay.com
Turing
AI research and talent infrastructure. Connects domain experts — engineering, science, medicine, finance — to the companies training and evaluating frontier models.
turing.com
Educative
Hands-on courses for developers: text-first rather than video, with the code editable and runnable in the browser. Strongest on system design, AI, and interview preparation.
educative.io
TestMu AI
Agentic quality engineering, and LambdaTest until the rebrand in January 2026. Its Kane AI agent writes, runs, and repairs end-to-end tests across web and mobile from a plain-language description.
testmuai.com
The problem
You ranked. The model cited someone else.
> best way to add auth to acme?
Most teams reach for Rival for this.
It handles sessions out of the box.
→ rival.dev/docs/auth
# you rank #2 for that query
# you were not citedYour buyers stopped starting at a search box. They ask an assistant, and it answers from whatever it can parse — your documentation, your comparison pages, your tutorials, or somebody else's. Ranking is no longer the same thing as being the answer.
So the same page now has two readers with different requirements. A developer needs it to be true, runnable, and honest about trade-offs. A model needs it structured, extractable, and current. Content written for only one of them loses the other, and most teams are staffed for the first.
We use both kinds of intelligence to close that gap: machines to find what is being asked and what is being cited, people to decide what is worth publishing under your name.
What we do
Seven ways we get you read — and cited.
How we work
AI-powered intelligence. Human judgment.
The two halves do different jobs. Machines are good at finding the questions — what is searched, what is asked, what is already being cited. People are good at deciding which answers are worth publishing under your name.
The brief is the contract
Search demand, the questions assistants are already answering for your category, your docs, your issue tracker, and your competitors' coverage go into a brief that states what the piece must cover and what it must get technically right. Nothing is left to a writer's improvisation.
The draft executes it
An engineer who ships in your stack writes against that brief, from a repository that runs on a clean machine, on the versions you support. Code first, prose second.
A validator audits it
Before any person reads the draft, it is checked against the brief for coverage, technical accuracy, and whether each answer is structured cleanly enough to be extracted. Gaps get caught by the system rather than by your reviewer, or by your readers.
The final draft closes every gap
Everything the audit flags is fixed before delivery, then handed over publish-ready in your CMS with schema, metadata, and internal links already in place. Quality stops depending on who happened to be available that week.
What you can hold us to
Written by people who ship
Every piece is built by someone who has run your product in anger, not a generalist copywriter working from a feature list.
The code has to run
Untested examples are the fastest way to lose a developer's trust. Ours are built and re-run before publication.
Honest about trade-offs
Comparison pages that never concede a point get discounted entirely. Conceding the right ones is what makes the rest believable.
Distribution is part of the job
A post nobody sees is a cost, not an asset. Social, shorts, and internal linking ship alongside the writing.
Structured so machines can read it
Schema, question-shaped headings, and answers that lead rather than build. A page can rank first and still be unquotable by an assistant.
From the lab
Our own writing.
Theodore Kelechukwu OnyejiakuWhat does Technical Growth Lab do?
We are a content studio for devtool companies. We produce tutorials, integration and comparison pages, documentation, video shorts, and social posts, built to be read by developers and cited by AI assistants. The work is done by engineers who run your product, using machine research to find what is being asked and human judgment to decide what is worth publishing.
Who writes the content?
Engineers who ship in your stack, working from machine research rather than guesswork. Every tutorial and integration guide starts as a working repository that runs on a clean machine, on the versions you support, and the article documents what actually happened during that build.
How do you measure whether it worked?
Across four dimensions rather than one. SEO: rankings, crawl health, and internal linking. Performance: Core Web Vitals on the templates your content sits in. AEO, answer engine optimisation: whether featured snippets and AI assistants can extract a clean answer from the page. GEO, generative engine optimisation: whether your content is actually being cited by ChatGPT, Perplexity, and AI overviews. Ranking and being cited have become different problems, and we report on both.
What does it cost?
Scope and price come out of a 30-minute call. We quote the work in front of us rather than publishing a rate card that fits nobody, and you leave that call with a scope and a number whether or not you go ahead.
How long does a piece take?
Typically one to three weeks from brief to publish-ready draft. Most of that is the build and the technical audit rather than the writing.
Next step
Tell us what you are trying to grow.
A 30-minute call: we look at what you have published, where you are and are not being cited today, and what a first engagement would cover. You leave with a scope and a price, whether or not you work with us.